Transformación en otras categorías de más amplio alcan-

The downwind turbine was erected at the Stellenbosch University Mariendahl small wind turbine testing facility as shown in Figure 5.15. The dowinwind turbine nacelle sits atop a 18 m tower and an addition 2 m extension which brings the hub height to ± 20.5 m. The power output from the wind turbine is measured using a Lovato DMG-700 Digital Multimeter and recorded (logged) using an EXP 1010 USB expansion module. The data is stored on a desktop computer situated on-site.

Figure 5.16 shows the average wind speed at the Mariendahl testing site over the weekend of 04 November 2017 to 06 November 2017. The wind speed at the testing facility is monitored using a Wind Monitor-JR Model 04101 wind speed and wind direction sensor. The wind speed is recorded using CR-300 datalogger from Campbell Scientific. From Figure 5.16, the average wind speed indicates that the wind turbine testing facility has strong wind conditions during the afternoon periods. The wind speed tends to level off in the evenings before retuning to average speeds between 4 m/s and 7 m/s.

Unfortunately, no meaningful wind turbine power data could be collected due to various time constraints which includes a robbery of the pre-existing on-site measuring equipment and power cables. However, much of the equipment has now been replaced and further wind turbine testing can now be conducted at the testing facility. Additional problems arose during the initial speed control tests conducted on-site. The downwind turbine nacelle and 2 m tower extension exhibits oscillatory behaviour in the presence of strong winds while under speed control (prior to grid connection). However, further field-testing showed that at low wind speeds, the wind turbine’s speed can be controlled at rated speed without any turbine oscillations. It was concluded that the severity of the oscillations is shown to be proportional to the speed of the wind at any given moment. A further issue is in the event where the downwind turbine blades are facing into the wind (the wrong direction). The 1.9 m blades aren’t always able to redirect the nacelle in the correct direction and as a result, no thrust is produced on the blades and the downwind turbine remains idle. Without a yaw drive, the downwind turbine is unable to turn itself in a direction where the blades are facing ”downwind” and consequently, no drivetrain acceleration is achieved.

18 m

2 m

Figure 5.15: 2.2 kW downwind turbine at the Stellenbosch University Mariendahl small wind turbine testing facility. 0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 W in d S p e e d ( m /s ) Time (hours)

Mariendahl Wind Speed over the period 04/11/2017* to 06/11/2017

04/11/2017 00:01:00

Figure 5.16: Average wind speed measurements recorded at the Mariendahl wind turbine testing facility over the weekend of 04 November 2017 to 06 November 2017.

5.6

Chapter Summary

In this chapter, the downwind turbine drivetrain’s dynamic performance was tested and evaluated in a laboratory setup. The downwind drivetrain (consisting of a gearbox, S-PMC and PMSG) was placed on a mechanical testbench and the turbine blades were replaced by an Allen-Bradley powerflex 700 variable speed drive (VSD) and an induction motor (IM). The VSD-IM combination was used as a means to provide a source of torque to the drivetrain in order to simulated the torque that conventional turbine blades would provide in the presence of wind.

A dynamic-step test was performed in order to identify the step-response of the downwind drivetrain. During this test, the grid connection controller (GCC) was placed in ”manual” mode where the duty cycle to the dumpload chopper circuit was manually stepped in order to record the speed response of the drivetrain when subjected to a 0.7 p.u. (100 Nm) input torque condition. The resulting speed-step response matched results of the simulated dynamic-step test. Consequently, the physical drivetrain’s

step response yielded identical process constants as those determined in Chapter 3. The subsequent process gain (kp), process time constant (τp) and process dead time (τd) were used to develop a digital

PI-controller using the tuning rules as outlined by the Internal Model Control (IMC) method.

The digital PI-controller was used in conjunction with the GCC and dumpload chopper circuit to control and maintain the downwind drivetrain speed at synchronous speed regardless of the input torque conditions. The PI-controller was able to control and maintain the drivetrain speed within the necessary boundaries as set out in [37] for torque inputs up until 100 Nm. However, the speed controller failed to maintain the drivetrain speed within the set boundaries for torque inputs greater than what was used in the dynamic step-test.

The GCC’s ability to connect the downwind drivetrain to the grid was also tested. The GCC was able to connect the drivetrain to the grid once all the synchronization conditions were met however the resulting grid-current transients proved to be a point of concern especially for torque inputs of 0.7 p.u. (100 Nm). The reason for the excessive grid current transients were shown (in Chapter 4) to be as a result of the acceleration of the generator prior to grid connection. It was shown that the GCC connected the drivetrain to the grid once the synchronization conditions were met but before the PI-controller had an opportunity to completely arrest the drivetrain’s acceleration.

Finally, the downwind drivetrain’s on-grid dynamic stability was evaluated. Two tests were conducted to determine what effect a 2.5 Hz and 8 Hz torque pulsation would have on the grid current. The 2.5 Hz and 8 Hz torque pulsation frequencies are as a result of wind sheer and tower shadow respectively. As was shown in Chapter 4, the downwind drivetrain was able to attenuate a 0.1 p.u. peak-to-peak torque pulsation component at a frequency of 8 Hz. Consequently, the tower shadow frequency had no effect on the grid current. Additionally, a 2.5 Hz, 0.1 p.u. peak-to-peak torque pulsation test was conducted in order to gauge and verify the bandwidth of the downwind drivetrain. The downwind dirvetrain was unable to attenuate the 0.1 p.u. torque pulsation at 2.5 Hz and the effects of the 2.5 Hz torque pulsation was evident on the grid current. Further testing suggested that the downwind drivetrain has a bandwidth of 4 Hz, as was shown in Chapter 4.

Chapter 6

Conclusion and Recommendations

The two main aims of this thesis were: (1) to design and develop a speed controller which has the ability to regulated the speed of a 2.2 kW downwind turbine at the synchronous speed required for grid connection; and (2) evaluate the dynamic performance of the downwind drivetrain.

6.1

Speed Control

In order to conduct speed control, an accurate simulation model of the downwind turbine drivetrain was formulated in MATLAB/Simulink. The downwind turbine simulation model was used to develop a PI-controller by simulating a dynamic step test in order to identify the drivetrain’s step response as well as its transfer function. From the step response, the relevant proportional (P) and integral (I) components of the PI-controller were derived using the internal model control (IMC) tuning rules. During laboratory testing, the same dynamic step-test was performed and the resulting drivetrain step response was shown to match the simulated results with a high degree of accuracy and as a result, the physical system proved to have the same transfer function as was found in simulation. Moreover, the resulting PI-controller was shown to provide acceptable performance in both the simulated and laboratory tests. It can therefore be concluded that the simulation methods and models used in this thesis provide a large degree of accuracy when it comes to predicating the response of the actual system. This is especially useful in the case of larger drivetrains where practical laboratory testing isn’t always possible.